| American and Canadian university-based English language programs use a variety of assessments to evaluate their students. The Standards for Educational and Psychological Testing direct test developers and users to validate the assessments they create or use. A variety of validation studies have been conducted on university-based English language tests. Many studies have focused on the nature and components of language and language acquisition. Some studies have explored tests' general quality, but few studies have investigated how language tests function over time. This study explores the growth characteristics (trajectories) of Michigan State University's English language placement listening and reading comprehension tests. The focus of this research is to investigate the inferences made with these tests over time and to identify variables that may effect students' growth trajectories.;Subjects used in this study attended Michigan State University's English Language Center from 1992 to 1996. A total of 308 students coming from 28 different countries are sampled. Students' sex, age, academic intent, nationality, and length of stay at the English Language Center are obtained and used in statistical analyses. Four separate analyses are conducted. First, both listening and reading comprehension tests are evaluated using classical test statistical procedures. Second, tests are calibrated and equated using a Rasch item response theory model. Third, equated person ability estimates for the listening and reading tests are combined with demographic information and analyzed descriptively. Finally, a growth model study using a two level hierarchical linear modeling (HLM) analysis is conducted. Two HLM growth models are investigated: linear and quadratic.;Both examinations exhibited satisfactory classical and IRT test statistical characteristics. A linear based HLM model fit both the reading and listening comprehension tests best. On the listening comprehension test, students staying more than one term at the English Language Center have much lower growth trajectories. Older students have significantly lower initial test scores compared to younger students. Likewise, undergraduate students tend to have higher initial listening scores, and certain Asian language groups have significantly lower initial listening scores. On the reading test, older students staying more than one term have significantly higher growth trajectories, and Japanese and Middle Eastern students have significantly lower initial reading test scores.;This study provides a glimpse of growth trends on English as a second language listening and reading comprehension tests used for placement and achievement purposes at an American university. This study serves as a starting point in the investigation of growth on language tests. The study also provides a view into the differences between growth trajectories on listening and reading comprehension tests. This work introduces the notion of examining a test's growth trajectory to determine how different test uses might be appropriate or inappropriate. |
